PROJECT 3 SUMMARY Immune checkpoint inhibitors have transformed melanoma treatment, producing durable responses, prolonged survival, and clinical benefit in a significant proportion of patients. Moreover, they delay recurrence and extend survival in the adjuvant melanoma setting, and have also shown efficacy in a range of different cancer types. However, immune checkpoint inhibition (ICI) therapy can also be accompanied by immune-related adverse events (irAEs) that impact multiple organs, cause significant morbidity, and require immunosuppression or discontinuation of ICI treatment. There is an urgent need to identify patients who will develop severe irAEs from ICI. This would enable us to optimize treatment selection and sequencing, justify preventive strategies to mitigate toxicity, and better manage toxicities. While there is intense interest in identifying markers to predict response to ICI, no pre-treatment biomarker tool can predict irAEs associated with ICI for any cancer type. The goal of our project is to develop a predictive tool that enables clinicians to minimize exposure of patients to severe toxicity, while maximizing clinical benefit from ICI. We hypothesize that a subset of melanoma patients has a baseline, sub-clinical autoimmune susceptibility, characterized by specific pre-existing autoantibodies (autoAbs) that can predict and exacerbate the development of toxicity from ICI therapy. We have identified autoAb signatures in baseline (pre-treatment) sera that predict severe immune toxicity in melanoma patients treated with ICI (AUC >0.95). Using a humanized mouse model, we found that autoAbs from baseline sera of melanoma patients can exacerbate irAEs from ICI. In this project, we propose to refine and validate baseline autoAb biomarker signatures of ICI toxicity using sera (n=600) from two large adjuvant ICI clinical trials for resected stage-III/IV melanoma (Aim 1). To understand the relevance of specific autoAbs to common irAEs (e.g., colitis) and to investigate an autoimmune predisposition in some patients, we will compare irAE-associated autoAbs with those from inflammatory bowel disease patients and from normal donors. We will use our humanized Fc?R mouse model to determine the cause-effect relationship between autoAbs and irAEs, with a focus on colitis, and for preclinical testing of prophylactic anti-TNF-? as a strategy to mitigate gastrointestinal (GI) toxicity from ICI (Aim 2). These findings will inform a biomarker-driven phase-II trial of prophylactic anti-TNF-? (infliximab) in patients receiving ICI therapy who are at high risk for developing severe diarrhea and colitis (Aim 3). Our work will inform personalized melanoma treatment strategies by validating a robust pre-treatment biomarker to enable clinicians to optimize ICI regimens and minimize patient exposure to severe irAEs. We will both identify an autoimmune susceptibility to irAE development and establish whether prophylactic TNF-? blockade mitigates development of GI toxicity from ICI in patients identified as being at high risk of these irAEs.